SDXL 1.0 VRAM requirements
The 2023 classic. One 6.9 GB checkpoint with the text encoders and VAE inside, and the biggest library of LoRAs and fine-tunes.
TypeImage
Parameters3.5B
Fits entirely from8 GB
8-bit or better from8 GB
01The files, and how much VRAM each needs
| File | Size | Needed | Min. VRAM | Quality | Source |
|---|---|---|---|---|---|
| 16-bit | 6.9 GB | 7.1 GB | 8 GB | the original weights | stabilityai/stable-diffusion-xl-base-1.0 → |
“Needed” = file + 1.2 GB working memory + 0.8 GB system reserve. The 6.94 GB checkpoint also holds the two text encoders and the VAE. During sampling only the UNet (about 5.1 GB at 16-bit, 2.6B parameters) has to sit in VRAM, so that is the figure used for the 16-bit file.
02By amount of VRAM
03Best GPU for SDXL
The cheapest cards (by launch price) that run it well, and every card sorted by memory: best GPU for SDXL → Planning bigger images or longer clips? Open the calculator →
04By graphics card
| GPU | VRAM | Verdict | Best file | Needed |
|---|---|---|---|---|
| Desktop graphics cards | ||||
| RTX 2060 6 GB | 6 GB | Offload only | 16-bit | 7.1 GB |
| RTX 3050 6 GB | 6 GB | Offload only | 16-bit | 7.1 GB |
| RTX 2070 Super 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 2080 Super 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3050 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3060 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3060 Ti 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3070 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3070 Ti 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4060 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4060 Ti 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5050 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5060 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5060 Ti 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 7600 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 9050 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 9060 XT 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| Arc B570 10 GB | 10 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3080 10 GB | 10 GB | Runs well | 16-bit | 7.1 GB |
| RTX 2080 Ti 11 GB | 11 GB | Runs well | 16-bit | 7.1 GB |
| Arc B580 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 2060 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3060 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3080 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3080 Ti 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4070 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4070 Super 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4070 Ti 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5070 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RX 7700 XT 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RX 9070 GRE 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| Arc A770 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4060 Ti 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4070 Ti Super 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4080 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4080 Super 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5060 Ti 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5070 Ti 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5080 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 7600 XT 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 7800 XT 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 7900 GRE 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 9060 XT 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 9070 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 9070 XT 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 7900 XT 20 GB | 20 GB | Runs well | 16-bit | 7.1 GB |
| Arc Pro B60 24 GB | 24 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3090 24 GB | 24 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3090 Ti 24 GB | 24 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4090 24 GB | 24 GB | Runs well | 16-bit | 7.1 GB |
| RX 7900 XTX 24 GB | 24 GB | Runs well | 16-bit | 7.1 GB |
| Arc Pro B70 32 GB | 32 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5090 32 GB | 32 GB | Runs well | 16-bit | 7.1 GB |
| Laptop GPUs | ||||
| RTX 3050 Laptop 4 GB | 4 GB | Offload only | 16-bit | 7.1 GB |
| RTX 3050 Ti Laptop 4 GB | 4 GB | Offload only | 16-bit | 7.1 GB |
| RTX 2060 Laptop 6 GB | 6 GB | Offload only | 16-bit | 7.1 GB |
| RTX 3050 Laptop 6 GB | 6 GB | Offload only | 16-bit | 7.1 GB |
| RTX 3060 Laptop 6 GB | 6 GB | Offload only | 16-bit | 7.1 GB |
| RTX 4050 Laptop 6 GB | 6 GB | Offload only | 16-bit | 7.1 GB |
| RTX 2070 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 2070 Super Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 2080 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 2080 Super Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3070 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3070 Ti Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3080 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4060 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4070 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5050 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5060 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5070 Laptop 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 7600M 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 7600M XT 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 7600S 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RX 7700S 8 GB | 8 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4080 Laptop 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5070 Laptop 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5070 Ti Laptop 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RX 7800M 12 GB | 12 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3080 Laptop 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 3080 Ti Laptop 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 4090 Laptop 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5080 Laptop 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RX 7900M 16 GB | 16 GB | Runs well | 16-bit | 7.1 GB |
| RTX 5090 Laptop 24 GB | 24 GB | Runs well | 16-bit | 7.1 GB |
| Unified memory | ||||
| Radeon 8060S (Strix Halo) 96 GB | 96 GB | Runs well | 16-bit | 7.1 GB |
| Radeon 8065S (Gorgon Halo) 160 GB | 160 GB | Runs well | 16-bit | 7.1 GB |
| Apple Silicon Macs (by memory) | ||||
| Mac 16 GB | 12.7 GB | Runs well | 16-bit | 7.1 GB |
| Mac 18 GB | 14.4 GB | Runs well | 16-bit | 7.1 GB |
| Mac 24 GB | 19.6 GB | Runs well | 16-bit | 7.1 GB |
| Mac 32 GB | 26.8 GB | Runs well | 16-bit | 7.1 GB |
| Mac 36 GB | 30.2 GB | Runs well | 16-bit | 7.1 GB |
| Mac 48 GB | 40.2 GB | Runs well | 16-bit | 7.1 GB |
| Mac 64 GB | 55.7 GB | Runs well | 16-bit | 7.1 GB |
| Mac 96 GB | 85 GB | Runs well | 16-bit | 7.1 GB |
| Mac 128 GB | 115.4 GB | Runs well | 16-bit | 7.1 GB |
| Mac 192 GB | 175.4 GB | Runs well | 16-bit | 7.1 GB |
| Mac 256 GB | 236.9 GB | Runs well | 16-bit | 7.1 GB |
| Mac 512 GB | 498.1 GB | Runs well | 16-bit | 7.1 GB |
05Text encoder, VAE and other files
The text encoder is inside the checkpoint, so there is nothing extra to download.
06Where the files go in ComfyUI
| File | Folder | Loader node |
|---|---|---|
| The checkpoint (.safetensors) | ComfyUI/models/checkpoints | Load Checkpoint |
07Measured and reported results
| Label | GPU | Setup | Result | Peak VRAM | Date | Source |
|---|---|---|---|---|---|---|
| reported | Arc A770 16 GB | SDXL 1.0 base · 1024x1024 · 20 steps “20/20 [00:14<00:00, 1.36it/s] ... Prompt executed in 23.44 seconds” ACER A770 16GB; thread benchmark = SDXL 1024x1024 20 steps seed 1; later runs 1.53 it/s / 14.28 s | 23.44 s / image · 1.36 it/s | — | 2025-07-10 | github.com → |
| reported | Arc B580 12 GB | sd_xl_base_1.0 · 1024x1024 · 20 steps “100%|██| 20/20 [00:05<00:00, 3.96it/s]” ComfyUI 'GPU Benchmark' thread: default workflow, SDXL 1.0 base, 1024x1024, seed 1, second run; Intel B580 Steel Legend OC 12 GB | 3.96 it/s | — | 2025-05-12 | github.com → |
| reported | RTX 2070 Laptop 8 GB | SDXL 1.0 base · 1024x1024 · 20 steps “2070 rtx mobile 20/20 [00:19<00:00, 1.05it/s] ... Prompt executed in 24.22 seconds” Thread benchmark SDXL 1024x1024 20 steps | 24.22 s / image · 1.05 it/s | — | 2024-03-04 | github.com → |
| reported | RTX 3070 Laptop 8 GB | SDXL 1.0 base · 1024x1024 · 20 steps “20/20 [00:11<00:00, 1.73it/s] ... Prompt executed in 15.44 seconds” RTX 3070 Laptop GPU, Asus ZenBook Duo; thread benchmark SDXL 1024x1024 20 steps | 15.44 s / image · 1.73 it/s | — | 2024-04-22 | github.com → |
| reported | RTX 3080 Ti 12 GB | sd_xl_base_1.0 · 1024x1024 · 20 steps “20/20 [00:04<00:00, 4.01it/s] Prompt executed in 5.60 second” ComfyUI 'GPU Benchmark' thread: default workflow, SDXL 1.0 base, 1024x1024, seed 1, second run | 5.6 s / image · 4.01 it/s | — | 2025-04-30 | github.com → |
| reported | RTX 3090 24 GB | sd_xl_base_1.0 · 1024x1024 “Prompt executed in 6.16 seconds” ComfyUI 'GPU Benchmark' thread: default workflow, SDXL 1.0 base, 1024x1024, seed 1, second run | 6.16 s / image | — | 2024-03-06 | github.com → |
| reported | RTX 4070 12 GB | sd_xl_base_1.0 · 1024x1024 · 20 steps “20/20 [00:06<00:00, 3.21it/s] Prompt executed in 7.13 seconds” ComfyUI 'GPU Benchmark' thread: default workflow, SDXL 1.0 base, 1024x1024, seed 1, second run; GPU stated as 'RTX 4070 12Gb' | 7.13 s / image · 3.21 it/s | — | 2024-03-21 | github.com → |
| reported | RX 7600 XT 16 GB | SDXL 1.0 base · 1024x1024 · 20 steps “20/20 [00:17<00:00, 1.13it/s] Prompt executed in 20.52 seconds” ComfyUI 0.3.29 Zluda, Windows 11, driver 25.3.1; thread benchmark SDXL 1024x1024 20 steps | 20.52 s / image · 1.13 it/s | — | 2025-04-18 | github.com → |
| reported | RX 9070 XT 16 GB | SDXL · 1024x1024 · 20 steps “| SDXL | 20 | 1024x1024 | 1.5it/s | 27.76s | Manual tiled VAE decoder |” Ubuntu 24.04, ROCm 6.4.1, PyTorch nightly, TUNABLEOP + AOTriton experimental, --use-pytorch-cross-attention; without tiled VAE 1.49 it/s / 34.56 s; default VAE decode could OOM | 27.76 s / image · 1.5 it/s | — | 2025-05-30 | github.com → |
Other people's numbers are copied as published, with a link. Send yours.
08Training a LoRA for SDXL
| Trainer | VRAM | Type | Settings and quote | Source |
|---|---|---|---|---|
| OneTrainer | 5.8 GB | example run | Community test (efhosci): fp8 weights, 1024px, batch 1, latent caching, SDP, gradient checkpointing; ~8.6 GB with fp16 weights “with resolution set to 1024 the VRAM usage peaked around 5.8 GB” | github.com → |
| sd-scripts | 8 GB | stated minimum | 1024px default; U-Net only, gradient checkpointing, cached TE outputs + latents, 8-bit optimizer or Adafactor, dim 4-8 for 8GB; 10GB recommended “The LoRA training can be done with 8GB GPU memory (10GB recommended).” | github.com → |
reported From each trainer's own docs, 2026-09-25. Why the figures differ →
09Every file tracked for SDXL
| File | Type | Size | Repo |
|---|---|---|---|
| sd_xl_base_1.0.safetensors | 16-bit | 6.94 GB | stabilityai/stable-diffusion-xl-base-1.0 → all-in-one checkpoint: includes text encoder(s) and/or VAE, not diffusion-model-only (UNet 2.6B + CLIP-L + OpenCLIP-G + VAE) |
Single checkpoint file loaded with Load Checkpoint. params_b is total incl. text encoders (UNet alone ~2.6B).